Memory In Memory: A Predictive Neural Network for Learning Higher-Order Non-Stationarity from Spatiotemporal Dynamics
arXiv:1811.07490
Abstract
Natural spatiotemporal processes can be highly non-stationary in many ways, e.g. the low-level non-stationarity such as spatial correlations or temporal dependencies of local pixel values; and the high-level variations such as the accumulation, deformation or dissipation of radar echoes in precipitation forecasting. From Cramer's Decomposition, any non-stationary process can be decomposed into deterministic, time-variant polynomials, plus a zero-mean stochastic term. By applying differencing operations appropriately, we may turn time-variant polynomials into a constant, making the deterministic component predictable. However, most previous recurrent neural networks for spatiotemporal prediction do not use the differential signals effectively, and their relatively simple state transition functions prevent them from learning too complicated variations in spacetime. We propose the Memory In Memory (MIM) networks and corresponding recurrent blocks for this purpose. The MIM blocks exploit the differential signals between adjacent recurrent states to model the non-stationary and approximately stationary properties in spatiotemporal dynamics with two cascaded, self-renewed memory modules. By stacking multiple MIM blocks, we could potentially handle higher-order non-stationarity. The MIM networks achieve the state-of-the-art results on four spatiotemporal prediction tasks across both synthetic and real-world datasets. We believe that the general idea of this work can be potentially applied to other time-series forecasting tasks.
References in corpus (13)
- Batch Normalization: Accelerating Deep Network Training by Reducing Internal Covariate Shift
- Sequence to Sequence Learning with Neural Networks
- Convolutional LSTM Network: A Machine Learning Approach for Precipitation Nowcasting
- Unsupervised Learning of Video Representations using LSTMs
- Scheduled Sampling for Sequence Prediction with Recurrent Neural Networks
- Decomposing Motion and Content for Natural Video Sequence Prediction
- PredRNN++: Towards A Resolution of the Deep-in-Time Dilemma in Spatiotemporal Predictive Learning
- Stochastic Video Generation with a Learned Prior
- Video (language) modeling: a baseline for generative models of natural videos
- Spatio-temporal video autoencoder with differentiable memory
- Stochastic Adversarial Video Prediction
- Visual Dynamics: Probabilistic Future Frame Synthesis via Cross Convolutional Networks
- Hierarchical Long-term Video Prediction without Supervision